Upload
dmilec
View
216
Download
0
Embed Size (px)
Citation preview
8/7/2019 Active_Crossovers_and_Filters
1/8
www.tubecad.com Copyright 2001 GlassWare All Rights Reserved NEXT >< PREVIOUS
>
1Pg.
What are active crossovers? In short, theyelectronic circuits that divide the audio spectrum
up into discrete bands of frequencies and theyfunction in place the passive crossovers found in
loudspeakers. The underlying motivation for theswitch to an active crossover is the improved
accuracy and flexibility of the active crossover
holds over the passive crossover.Loudspeakers represent truly complex
impedances, which only an equally complexpassive crossovers can match. Thus deriving the
required crossover frequency and slope isdifficult with a passive crossover and changing a preexisting passive crossover frequency is
anything but easy. Active crossovers, on the
other hand, remove the large passivecomponents that make up the passive crossover.
This is for the good, as the hundred feet ofmagnet wire that makes up the inductors and the
two back-to-back electrolytic capacitors thatmake up most crossover non-polarized
capacitors are not missed: these components are
far from ideal. And the power amplifier, oncefreed from having to work through this dreck,exercises a better control of the loudspeaker
drivers. For example, a damping factor of 100
means little, if the passive crossover adds 1 ohmof DC resistance to the mix, thereby decreasing
the effective damping ratio to 8.Active crossovers also allow for a frequency
tailoring that would be altogether impossible orat least incur efficiency penalties with a passive
crossover. For example, with a network
designed by Linkwitz, we can effectively shiftthe resonant frequency and Q of a loudspeakerdriver. Furthermore, a high Q crossover can
boost a drooping low frequency response of alow Q speaker while filtering away sub-sonic
garbage, such as record warp.
Bi-Ampping Psycho-AcousticsAn added advantage is a psycho-acoustic
phenomena wherein bi-ampped systems seem
more powerful than the sum of the amplifier
wattages, e.g. two 36 watt amplifiers soundmuch more powerful than a 72 watt amplifier.
How is that possible? Half of the answer lies
in the output voltage adding together rather thanthe wattages adding together. Wattage is basedon the voltage squared. For example, 24 peak
volts of output signal equals 36 RMS watts, but48 peak volts of output signal equals 144 RMS
watts. In other words, doubling the voltagequadruples the wattage; tripling, increases the
wattage by ninefold.
Active Crossovers and Filters
Let's imagine a crossover point of 500 Hz. If a
144 watt amplifier is presented with a signal oftwo 24 volt tones, say 100 Hz and 2 kHz, the
signal will trace a 2 kHz sine wavesuperimposed on a 100 Hz wave. The total peak
voltage is 48 volts, which equals 144 RMSwatts. Remember that a given instant the
amplifiers output is at only one specific voltage.(The plate cannot be at +24 and at 12 volts at
the same time. We are dealing with cooper wireand voltage, not fiber optics and light.) A
passive crossover splits these two frequencies
from the output from a 144 watt amplifier intotwo 24 volt signals. The active crossover alsoseparates the two frequencies so that two 36 watt
amplifiers can play these two tones at the samevolume as the single 144 watt amplifier can, i.e.
24 volts at 100 Hz and 24 volts at 2 kHz.
2 kHz200 Hz20 Hz 20 kHz
+ =200 Hz
2 kHz
200 Hz + 2 kHz added together
8/7/2019 Active_Crossovers_and_Filters
2/8
www.tubecad.com Copyright 2001 GlassWare All Rights Reserved NEXT >< PREVIOUS
>
2Pg.
Of course, if both tones had been at
frequencies below the 500 Hz crossover point,say 100 and 300 Hz, one 36 watt amplifier
would clip its output while the second amplifiersat idle. While this seems to limit the power
increasing effect, this indirectly leads to secondpower increasing effect.
One last argument in favor of active
crossovers: headphone listening can be superb, but it suffers from the lack of a visceral
component; the floor just doesnt shake withheadphonesbut it can. For over two decades
my preferred way of listening to headphones has been to use them with powered subwoofers.
With a high frequency limit of 80 Hz thesubwoofers seldom even move with pop music,
but with Mahlers 3rd symphony, they workovertime. (The missing octave is filled in and yet
your neighbors have a hard time locating you as
the source of the low growls, as the ear has ahard time localizing ultra-low frequencies.)
Filter TypesWhat is the difference between a crossover
and filter? A crossover is specifically meant to
work with loudspeakers and it is usually madeup of several filters. In other words, a filter is the
more general term, which suits a circuit that hassuch a wide application. Almost all electronic
devices employ some filters. The radio and TV
set are filters of sorts, as they select only anarrow band of frequencies at a time. A filter
can be made out of as little a single capacitor orinductor (and a load resistance). A filter can also
consist of twenty resistors and twenty capacitorsand many amplifiers.
FiltersAnalog Digital
Passive Active
Lowpass Lowpass
Bandpass Bandpass
Notch Notch
Highpass Highpass
+ =100 Hz
300 Hz
The ear is most sensitive at about 3 kHz
(roughly, the resonate frequency of the earcanal). As the frequency falls from this
frequency, so does our sensitivity to it. So if a bi-ampped system sees a signal that only clips
the woofer amplifier, the ear will not complainas much as it would if the tweeter amplifier had
clipped. A further reason is that woofers areslow heavy devices that cannot reproduce the
harsh high harmonic of the clipped lowfrequency waveform. So while a low frequency
amplifier may clip repeatedly, the woofer willnot readily betray that fact. And since it isusually the low frequencies the demand the
greatest share of power, the two 36 wattamplifiers may even seem more powerful than a
144 watt amplifier as long as the high frequencyamplifier has not clipped. (This might helpexplain why a 3 watt tube SE amplifier can
sound as powerful as a 30 watt solid-state
amplifier: the ear hears clipping as the limit to power. Transformer coupled, feedback free
amplifiers have the softest clippingcharacteristics of any amplifier, as thetransformers limit the transfer of highfrequencies and the absence of feedback means
the driver stage will not try force a sharper slopeon the output tubes flattened waveform.)
100 Hz + 300 Hz added together and clipped
In all filters, however, some frequencies are
allowed to pass, while others are attenuated. The portion that is allowed to pass is named the passband and the portion that is excluded is
named the stopband and the portion of overlap between these two bands is named the
transitional-band.
8/7/2019 Active_Crossovers_and_Filters
3/8
www.tubecad.com Copyright 2001 GlassWare All Rights Reserved NEXT >< PREVIOUS
>
3Pg.
If only low frequencies are passed, we call this
filter a low-pass filter, as that is what it passes. If
only high frequencies are passed, we call thisfilter a high-pass filter. If only a band of
frequencies, say 500 to 5000 Hz, are passed, wecall this filter a band-pass filter. If all frequency
are passed save for a very narrow band offrequencies, we call this filter a notch filter.
Beyond these four basic divisions, a furtherclassification is applied based the steepness ofthe rejection of undesired frequencies.
Since the slopes steepness is marked by the
order in a filters design. (The number of poles
is also sometimes used as a short handdescription of the steepness.) For example, a
single order filter, also called a single pole filter,has a crossover slope of -20 dB per decade,which equals -6 dB per octave. Thus a second
order filter attenuates at 12 dB per octave; athird order filter, -18 dB per octave; a fourth
order filter, -24 dB per octave; a fifth orderfilter, -30 dB per octave; and a sixth order
filter, -36 dB per octave.
The most common filter alignments are the
Butterworth and the Bessel (AKA Thomson).Less common alignments are the Gaussian,
Chebyshev and the elliptic (AKA Cauer). Arecent addition to the palette of filters is the
Linkwitz-Riley alignment (AKA Butterworthsquared), which was specifically designed for
use as loudspeakers crossover (More on this typeof filter to follow.).
Common Audio Filter TypesThe Butterworth filter is the most commonly
used filter type in audio applications. It has afairly flat time response, a fairly crisp transition
shape and the flattest passband response. Incontrast, the Bessel filter offers a flatter time
response, but not as sharp a transition shape orflat a passband response. The Chebyshev and the
elliptic filter are seldom used in everyday audiogear, as these filters bring too many undesirable
characteristics such as ripples in the passband orwild phase shifts.
The Linkwitz-Riley filter improves on the
Butterworth when feeding loudspeakers. Oddorder crossovers (6 dB and 18 dB per octave
filters) are marked by of 90 degree multiples in phase differences between outputs. The sum of
two 90 degree equal amplitude signals is +3 dB boost in output. Thus the 3 dB point of the
filter works to yield a flat transition region.Even-order crossovers (12 dB and 24 dB per
octave filters) are marked by 180 degreemultiples of phase differences between outputs.
The sum of two 180 degree phase shifted, butequal amplitude, signals is a deep null in the
output. The solution is to reverse the phase ofone of the loudspeaker drivers to eliminate the
phase difference at the crossover frequency.However, when two loudspeakers play the exact
same signal (no amplitude or phase differences),
the result is a twofold increase (+6 dB boost) involume. So if we wish to reconstruct the outputsof one lowpass and one highpass filter, we must
tailor the attenuation at the crossover frequency
to match the intended summing device.
Lowpass Bandpass HighpassNotch
F 10F0
-20
-3
1st2nd3rd4th
The next taxonomic division is made based onthe filters alignment. The alignment is based on
the relative position of the poles in the filter,which in turn gives rise to the different shapes
each filter exhibits across its frequency range.
Some filters offer a sharp transition from passband to stopband, but at the cost of some
ripple and even ringing in the frequencyresponse. Other filters offer a soft transition that
droops at the transition frequency that is ripplefree. In short, an analog filters design hopes to
optimize one or two parameters, but always atthe cost of some other parameters.
8/7/2019 Active_Crossovers_and_Filters
4/8
www.tubecad.com Copyright 2001 GlassWare All Rights Reserved NEXT >< PREVIOUS
>
4Pg.
Thus, when using loudspeakers that see the
same phase and amplitude signal, we cannot usea 3 dB down point, as it would yield +3 dB;
thus, we need a 6 dB attenuation at thecrossover frequency in order to ensure a flat
frequency response. The same phase is the keypoint here. If a Butterworth second or fourth or
sixth order crossover is used, the frequencyresponse will display a bump at the crossover
frequency. On the other hand, using a Linkwitz-Riley 2nd or fourth order crossover will yield a
flat frequency response at the crossover
frequency.Well at least that is the theory. Complications
arise: the distance between drivers, thefrequency response of each driver, the front-to- back spacing of the drivers voicecoils. If the
distance between drivers exceeds the wavelength
of the crossover frequency, or if one or bothdrivers droop at the crossover frequency, or if
the tweeter acoustic center is substantially infront of the woofers, then the Butterworth
aligned crossover might actually prove flatter.Do not falsely imagine that an even order
crossover is phase flat; it isnt. The actual phaserelation between 2nd order lowpass and
highpass filters is a constant 180 degrees phasedifference, which when one driver's connection
is inverted, yields a constant phase relation
between drivers, but not a flat phase response.
A Phase Flat LoudspeakerOne interesting loudspeaker configuration
was created by Philips. An ostensibly two-wayloudspeaker was designed using a second order
Butterworth crossover, but without the tweetersphase reversal. Yes this resulted in a deep suck-
out. The suck-out was then filled in by using a bandpass filter (-6 dB slopes) feeding a full-
range driver. This extra speaker saw only the
crossover frequency unattenuated, as all thefrequencies below or above this frequency were
attenuated at 6 dB per octave. The sum of allthree drivers output equaled a phase-coherent
flat frequency response. In this speaker system,the woofer was prevented from going too high,
while the tweeter is protected from lowfrequencies, while fullrange driver filled in the
hole. Unfortunately, this solution, like mostspeaker innovations, disappeared long ago.
0-3
Crossover Testing
Evaluating active crossovers under actual use
is difficult. Which component is making thebiggest difference? The active crossover itself or
the extra amplifiers and speaker cables? Alistening test I have used is to take a pair of
fullrange conventional loudspeakers and applythe crossover under test using my present
amplifier and cables. This means there is noworry about relative efficiencies and frequency
response limitations.
The procedure is simple enough; place onespeaker on top of the other and set the crossover
frequency to some value between 300-700 Hz(too high a frequency will result in lobbingeffects due to greater than wavelength spacing
between drivers). This test requires a monosignal source and I recommend any of the mono
Verve recordings of Ella Fitzgerald.
Phase differences between 2nd order Butterworthhigh-pass filter (white line) and its low-passcomplement (yellow line)
Fc/10 Fc 10xFc
180
0
-180
8/7/2019 Active_Crossovers_and_Filters
5/8
www.tubecad.com Copyright 2001 GlassWare All Rights Reserved NEXT >< PREVIOUS
>
5Pg.
General crossover test setup
500 Hz
500 Hz
First listen to just the loudspeaker on top
(fullrange, without the active crossover in otherwords). It takes some time to get used to mono,
so play about 30 minutes worth. Next introduce
the active crossover and feed the highpass signalto the top loudspeaker. Now listen to the same
30 minutes of music.
Loudspeaker diaphragms move in response to
signals and the large diaphragm movements
caused by low frequencies raise the highfrequencies while the drivers cone movestoward us and then lower the high frequencieswhen the cone moves away from us.
Consequently, in absolute terms, it is impossiblefor anyone driver to reproduce more than one
pure tone at a time. This is one reason why
electrostatic and planer speakers sound as goodas they do; the large surface areas do not need to
move very far to produce high sound levels.Three-way, four-way, five-way, and six-way
loudspeaker also greatly reduce this distortion,while admittedly adding huge crossover
problems. And the worst offenders are smallfullrange single driver based loudspeakers.
Bi-Ampped Identical LoudspeakersI once heard something like the test setup with
four large Advent loudspeaker from the early80s. A friend bought four speakers for
quadraphonic use and instead used all four in astereophonic setup with a active crossover set at
100 Hz. The improvement in midrange clarityand bass definition was markedly better. In fact,
it sounded much better than it should seempossible. (A similar experience occurred when I
heard a speaker setup that consisted of eight highquality mini-monitors configured as two
speakers. No crossover was used and fourspeaker enclosures were stacked on each other
per side, with only one facing the listener. The
resulting impedance was identical to that of asingle speaker, as they were wired in series-
parallel. Because each speaker saw half of theavailable output voltage from the power
amplifier, its response was down 6 dB; but as theradiating area had increased fourfold, the output
response gained +12 dB of gain, which in theend, yielded +6 dB of gain.)
If the crossover is working well, the soundshould be very similar to that from the single
loudspeaker. The usual result is that the sound ismuch better than that from the sole speaker.
Why? The power increasing effect is part of thereason. The rest of the answer is found in the
reduction in distortion that results from freeingthe top loudspeaker from the onerous task of
reproducing low frequencies. In other words,
the top loudspeaker is less likely to mechanicallyclip. Furthermore, there is a large reduction in aseldom mentioned distortion: frequency
modulation distortion, FMD.
FMDThis distortion is related to Doppler
distortion: the phenomenon of frequencies
increasing in pitch as a sound producing objectnears us and the sound decreasing in pitch as the
object retreats. The sound of a race car
approaching and passing is the usually givenexample. RRRRR R R R R R R R R R R.
8/7/2019 Active_Crossovers_and_Filters
6/8
www.tubecad.com Copyright 2001 GlassWare All Rights Reserved NEXT >< PREVIOUS
>
6Pg.
+
-
+
+
In same vein, if you are pleased with your
present loudspeakers, but wish that they provided a little more headroom or clarity at
high volumes, then this doubling ofloudspeakers might be the best way to go.
Active crossovers are certainly nice, but notstrictly necessary, as one choke and one
capacitor are all that is needed to make a passivesingle order crossover. My recommendation is to
select a crossover frequency that is thegeometric mean of the lowest and highest
frequencies produced by your loudspeakers. For
example, 600 Hz for speakers that extend downto 20 Hz and up to 20 kHz; 1400 Hz for that
only extend down to 100 Hz (mini-monitors).
Fgeometric = (Flow x Fhigh)I also recommend using the shunt arrangement
rather than the series, as it provides better
damping. (The shunt arrangement attenuates thestopband by shunting the driver with an everdecreasing impedance; where the series
arrangement attenuates the stopband by presenting the driver with an ever increasing
impedance.) Of course, with a passive crossover,none of the power increasing effects are
bestowed.
Choosing Crossover SlopesIn theory, nothing beats a simple 1st order
crossover. It results in a flat, time coherentfrequency response. Actual practice differs. The1st order filter provides a shallow attenuation
slope that is too gradual to protect fragile
tweeters from brutal low frequencies and toogradual to keep woofers from working up into
1st 3rd
100 Hz
their high frequency limits, as a two to three
octave overlap is recommended to make thisfilter work well. However, as the number of
crossover points increase, the more likely it isthat this gentle filter slope will provide sufficient
protection.On the other hand, if you are adding a powered
subwoofer to an existing fullrange loudspeaker,the 1st order highpass filter is a good choice for
the fullrange speaker and a 3rd order filter forthe subwoofer. This combination seems to work
well in terms of amplitude response. It also has a
minimalist aspect that is gratifying: thesubwoofer contributes only its low frequency
heft and then quickly gets out of the way of thelower midrange; the satellite loudspeaker sees avery gentle highpass slope that allows it extend
down into the deep bass as it very slowly falls
off in volume with decrease frequency.
Those using zero feedback tube amplifiers,
will find a preexisting 1st order highpass filter in
their power amplifiers. The coupling capacitorthat connects to the grids of the output tubesdefines a highpass filter that is usually set tosome very low frequency, such as 1 to 5 Hz.
Decreasing the value of this capacitor raises thecrossover frequency. By choosing the right
value, we can forgo the need for an external
crossover. The formula is an easy one:
C = 159155/Fc/R,
Where Fc equals the desired crossoverfrequency, C equals the value of the coupling
capacitor in F, and R equals the value of thegrid resistor. Truly minimalist. One trick I have
wanted to try for some time now is using asatellite loudspeaker (a sealed box with a Q of 1)
along with a 1st order filter, which comes with aQ of 0.707 (set to the Fo of the enclosure).
8/7/2019 Active_Crossovers_and_Filters
7/8
www.tubecad.com Copyright 2001 GlassWare All Rights Reserved NEXT >< PREVIOUS
>
7Pg.
The Qs are multiplied against each other,
which will give the loudspeaker an effective Qof 0.707. The advantage of this arrangement lies
in the smaller enclosure needed for a higher Qloudspeaker alignment and the natural 2nd order
highpass filter function of the sealed boxenclosure adding to 1st order filter to make a 3rd
order highpass filter with a -3 dip at thecrossover frequency.
To work best this 1st order filter must comebefore the power amplifier, as using a capacitor
in series with the loudspeaker will not work due
to the impedance spike at the speakersresonance. The subwoofer's lowpass filter should
be actively realized with an active 3rd orderButterworth filter. A further refinement would be to add an active 2nd order or 4th order
highpass filter whose cutoff frequency would
equal the Fs of the subwoofer. This would bothwork to protect the satellite speaker from
excessive low frequency cone excursions andwork to bring both subwoofer and satellite into
the same phase relationships, as the subwoofersown box resonance defines a highpass filter (2nd
order for sealed boxes and 4th order for bass-reflex enclosures.)
1st 3rd
100 Hz
2nd
30 Hz
3rd
For most drivers that suffer from either
limited power handling or limited frequencyresponse, sharper cutoffs are needed. The 2nd,
3rd, and 4th order slopes both protect and isolatethe drivers. Many audiophiles, however, distrust
higher order filters, fearing the signals phasereversals and complexity of the crossovers. Yet
they do not realize that the speakers themselves bring phase aberrations because of resonancesdriver breakup at higher frequencies. For
example, most tweeters have a resonantfrequency between 800 to 2 kHz and all cone
and dome drivers must flex once thecircumference of the diaphragm becomes largerthan the twice the wavelength of the frequency
being reproduced. (Strictly speaking, for
example, we should not let an 8 inch wooferextend beyond 1400 Hz; but we do so regularly.)
Here is where active filters shine. As has beenmentioned, the problem with using passivefilters with loudspeakers is that they seldomfunction entirely to plan and this poor
performance is due to the passive componentsfalling short of their ideals.
Inductors (chokes) are made up of longlengths of wire, which adds resistance. Capacitor
also carry an ESR (effective series resistance)
that upsets inter-relations within the filter. Andloudspeaker drivers are anything but pure
resistances. All of these failings add up to a filterthat does not match its textbook model. On the
other hand, the active filter can match itstextbook template easily. Additionally, active
filters can make do with only resistors andcapacitors to define their poles. Doing without
inductors is a relief in terms of money and space,as the lower the frequency, the bigger the
inductor. And because the active filter presents
an almost infinite input impedance to thefrequency tailoring parts, high valued resistors
can be used, which only require small valuedcapacitors to match. For example, an 8 ohm
loudspeaker driver crossing over at 1 kHzrequires a 20 F capacitor, whereas a 10k
resistor only needs a .0159 F capacitor. In otherwords, active filters present a win, win situation.
Using the speaker's own effective filters to completea crossover . Here the speaker's 2nd order functionis added to a 1st order filter to yield an effective 3rdorder slope.
8/7/2019 Active_Crossovers_and_Filters
8/8
www.tubecad.com Copyright 2001 GlassWare All Rights Reserved NEXT >< PREVIOUS
>
8Pg.
Mix and MatchActive filters can be cascaded to yield a
degree of frequency tailoring that is all butimpossible with passive filters. The passive filter
works best with known, fixed input andterminating impedances, which a cascading
passive filter cannot readily do.For example, two active 3rd order filters
cascaded equals a 6th order filter. An active
highpass and a lowpass filter cascaded can yieldbandpass filter; and these same filters fed into a
mixer can yield a notch filter.
ConclusionIn this article we have looked into the why
bother with active crossovers and filters? Wehave seen that active crossovers and active
filters in general offer some real advantages over
their passive brethren, as they are more accurate,flexible, and easy to implement. In the nextarticle, we will look into designing tube
crossovers.
//JRB
Suggested Readings
Active Filter Cookbook
By Don Lancaster, 1975, 95, 96
IC based active filters explained in detail. A
classic that is readily available new and used.
Audio/Radio Handbook
By National Semiconductor, 1980
This book has a chapter whimsically namedFloobydust, which covers some interesting
active crossovers, including asymmetrical filters.
Passive and Active Filters
By Wai-Kai Chen, 1986
The book to own, if you are serious about filters.
A classic, but not for those fearful of math.
Understanding Electronic Filters
By Owen Bishop, 1996This book is the best place for the novice to
start. Should be read from start to finish.
O v e r 1 0 ,0 0 0 sq u a r e f e e t o f :
Tubes, books, transformers, sockets, friendly
folks, capacitors, resistors, literature,
cabinet restoration materials, friendly folks,
wire, grill cloth, gifts, tools, information...
and did we mention the friendly folks?
Tubes
Kits & Power Supplies
SocketsGuitar Items
Transformers
Tools & Chemicals
Capacitors
Books
Other Parts & Supplies
Software
Restoration Products
Antique Electronic Supply
Get our 2001 catalog
absolutely free!
68 pages of products
and information (most
of which are alsoavailable on our Web
site). It's all here!